import warnings

warnings.filterwarnings("ignore")

import pandas as pd
import numpy as np
import datetime
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.preprocessing import OneHotEncoder
from sklearn.preprocessing import MinMaxScaler
from sklearn.model_selection import train_test_split
from sklearn.pipeline import Pipeline
from sklearn.linear_model import LinearRegression
from sklearn.model_selection import KFold, cross_validate
from sklearn.model_selection import cross_val_score
import time
from sklearn.model_selection import cross_val_predict
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import Lasso
from sklearn.metrics import r2_score
from sklearn.metrics import (
    make_scorer,
    mean_squared_log_error,
    mean_absolute_error,
    mean_squared_error,
)
from sklearn.metrics import (
    mean_squared_error,
    mean_absolute_percentage_error,
)
from sklearn.metrics import make_scorer
from sklearn.metrics import max_error

from sklearn.ensemble import RandomForestRegressor, AdaBoostRegressor, BaggingRegressor
from sklearn.svm import SVR
from sklearn.model_selection import TimeSeriesSplit
from sklearn.model_selection import GridSearchCV

df_bike = pd.read_csv("london_merged.csv")

df_bike.head()

df_bike.columns

Index(['timestamp', 'cnt', 't1', 't2', 'hum', 'wind_speed', 'weather_code',
       'is_holiday', 'is_weekend', 'season'],
      dtype='object')

df_bike.shape

(17414, 10)

print(
    "Die Gesamtzahl der Datensätze im Datensatz beträgt: "
    + str(df_bike.timestamp.count())
    + ", from: "
    + str(df_bike.timestamp.min())
    + " to: "
    + str(df_bike.timestamp.max())
)

Die Gesamtzahl der Datensätze im Datensatz beträgt: 17414, from: 2015-01-04 00:00:00 to: 2017-01-03 23:00:00

df_bike.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 17414 entries, 0 to 17413
Data columns (total 10 columns):
 #   Column        Non-Null Count  Dtype  
---  ------        --------------  -----  
 0   timestamp     17414 non-null  object 
 1   cnt           17414 non-null  int64  
 2   t1            17414 non-null  float64
 3   t2            17414 non-null  float64
 4   hum           17414 non-null  float64
 5   wind_speed    17414 non-null  float64
 6   weather_code  17414 non-null  float64
 7   is_holiday    17414 non-null  float64
 8   is_weekend    17414 non-null  float64
 9   season        17414 non-null  float64
dtypes: float64(8), int64(1), object(1)
memory usage: 1.3+ MB

df_bike.rename(
    columns={"cnt": "count", "t1": "temp", "t2": "temp_feels_like", "hum": "humidity"},
    inplace=True,
)

df_bike["timestamp"] = pd.to_datetime(df_bike["timestamp"])

df_bike["year"] = df_bike["timestamp"].dt.year

df_bike["month"] = df_bike["timestamp"].dt.month_name()

df_bike["time"] = pd.to_datetime(df_bike["timestamp"]).dt.time

df_bike["day_of_week"] = df_bike["timestamp"].dt.day_name()

# تحويل رموز الفصول إلى أسماء
season_dict = {0.0: "Spring", 1.0: "Summer", 2.0: "Autumn", 3.0: "Winter"}
df_bike["season_name"] = df_bike["season"].map(season_dict)

df_bike["day"] = df_bike["timestamp"].dt.date

df_bike.describe()

df_bike["timestamp"].value_counts()

timestamp
2015-01-04 00:00:00    1
2016-05-04 12:00:00    1
2016-05-04 18:00:00    1
2016-05-04 17:00:00    1
2016-05-04 16:00:00    1
                      ..
2015-09-03 17:00:00    1
2015-09-03 18:00:00    1
2015-09-03 19:00:00    1
2015-09-03 20:00:00    1
2017-01-03 23:00:00    1
Name: count, Length: 17414, dtype: int64

df_bike["is_weekend"].value_counts()

is_weekend
0.0    12444
1.0     4970
Name: count, dtype: int64

df_bike["season"].value_counts()

season
0.0    4394
1.0    4387
3.0    4330
2.0    4303
Name: count, dtype: int64

df_bike["weather_code"].value_counts()

weather_code
1.0     6150
2.0     4034
3.0     3551
7.0     2141
4.0     1464
26.0      60
10.0      14
Name: count, dtype: int64

df_bike["is_holiday"].value_counts()

is_holiday
0.0    17030
1.0      384
Name: count, dtype: int64

print("Feiertage pro Jahr: ~", 384 / 17414 * 365.25)

Feiertage pro Jahr: ~ 8.05420925691972

df_bike.index = pd.to_datetime(df_bike["timestamp"])

agg_freq = "1W"  # agg_freq = "1W": هذا يحدد التردد الذي سيتم استخدامه لتجميع البيانات. في هذه الحالة، “1W” يعني أن البيانات ستتم تجميعها بناءً على الأسبوع.

counts = (
    df_bike["count"].groupby(pd.Grouper(freq=agg_freq)).sum()
)  # عند استخدام pd.Grouper, يجب أن يكون الفهرس من نوع DatetimeIndex لأن pd.Grouper يقوم بتجميع البيانات بناءً على التواريخ
temps = (
    df_bike["temp"].groupby(pd.Grouper(freq=agg_freq)).mean()
)  # هذا يقوم بتجميع درجات الحرارة (“t1”) بناءً على التردد الذي حددناه ويحسب المتوسط لكل أسبوع.
hums = df_bike["humidity"].groupby(pd.Grouper(freq=agg_freq)).mean()
winds = df_bike["wind_speed"].groupby(pd.Grouper(freq=agg_freq)).mean()

fig, axes = plt.subplots(4, 1)
fig.set_figheight(12)
fig.set_figwidth(12)
axes[0].set_title("Bike shares")
sns.lineplot(x=counts.index, y=counts.values, ax=axes[0], color="green")
axes[1].set_title("Temperature")
sns.lineplot(x=temps.index, y=temps.values, ax=axes[1], color="green")
axes[2].set_title("Humidity")
sns.lineplot(x=hums.index, y=hums.values, ax=axes[2], color="green")
axes[3].set_title("Wind speed")
sns.lineplot(x=winds.index, y=winds.values, ax=axes[3], color="green")
plt.tight_layout()

mean_agg = (
    df_bike[["temp", "temp_feels_like", "humidity", "wind_speed", "season"]]
    .groupby(pd.Grouper(freq=agg_freq))
    .mean()
)

season_key = ["Spring", "Summer", "Autumn", "Winter"]
seasons = (
    mean_agg["season"]
    .apply(lambda x: int(np.floor(x)))
    .apply(
        lambda x: season_key[x]
    )  # هذا يقوم بتحويل قيم الموسم في mean_agg من أرقام إلى أسماء الفصول باستخدام season_key. يتم تخزين النتائج في seasons.
)
mean_agg["season"] = seasons

g = sns.PairGrid(mean_agg, hue="season")
g.map_diag(sns.histplot)
g.map_offdiag(sns.scatterplot)
g.add_legend()

<seaborn.axisgrid.PairGrid at 0x1fba013d2d0>

weather_counts = mean_agg[["temp", "humidity", "wind_speed", "season"]].join(counts)
weather_counts

g = sns.PairGrid(
    weather_counts,
    x_vars=[
        "temp",
        "humidity",
        "wind_speed",
    ],
    y_vars="count",
    hue="season",
    height=5,
    # palette="pastel",
)
g.map(sns.scatterplot)
g.add_legend()

<seaborn.axisgrid.PairGrid at 0x1fba4487d90>

df_bike = df_bike.reset_index(drop=True)

<matplotlib.lines.Line2D at 0x1fba4fc1650>

<matplotlib.lines.Line2D at 0x1fba7baf210>

<matplotlib.lines.Line2D at 0x1fba99d9650>

<matplotlib.lines.Line2D at 0x1fba9b82cd0>

<matplotlib.lines.Line2D at 0x1fba9d31050>

<matplotlib.lines.Line2D at 0x1fba9b21dd0>

timestamp          0
count              0
temp               0
temp_feels_like    0
humidity           0
wind_speed         0
weather_code       0
is_holiday         0
is_weekend         0
season             0
year               0
month              0
time               0
day_of_week        0
season_name        0
day                0
dtype: int64

time
07:00:00      2
08:00:00    273
09:00:00      1
13:00:00      3
14:00:00      3
15:00:00      6
16:00:00      5
17:00:00    120
18:00:00     99
19:00:00      2
dtype: int64

The mean of shared bikes per day is: 27268.45479452055

Text(0.5, 1.0, 'Correlation between attributes')

df_bike = df_bike.reset_index(drop=True)

# تعيين حجم الرسم والدقة
plt.figure(figsize=(20, 5), dpi=300)

# إنشاء الرسم البياني لتوزيع عدد ركوب الدراجات
sns.distplot(df_bike["count"])

# تعيين العنوان باللغة الألمانية
plt.title("Verteilung der Fahrradfahrten", fontsize=20)

# إزالة الحواف العلوية واليمنى من الرسم البياني
sns.despine(top=True)

# إضافة خط متوسط القيمة بشكل متقطع
mean_count = df_bike["count"].mean()
plt.axvline(mean_count, ls="--", color="red", label=f"Mittelwert: {mean_count:.2f}")
plt.legend()

# إظهار الرسم البياني
plt.show()

plt.figure(figsize=(20, 5), dpi=300)
sns.histplot(data=df_bike, x="temp", kde=True, color="skyblue")
plt.axvline(
    df_bike["temp"].mean(),
    ls="--",
    color="red",
    label=f'Durchschnittliche Temperatur: {df_bike["temp"].mean():.2f}',
)

# Festlegen des Titels und der Achsenbeschriftungen auf Deutsch
plt.title("Verteilung der tatsächlichen Temperatur", fontsize=6)
plt.xlabel("Tatsächliche Temperatur")
plt.ylabel("Anzahl der Datensätze")

# Anzeigen der Legende und des Diagramms
plt.legend()
plt.show()

plt.figure(figsize=(20, 5), dpi=300)
sns.histplot(data=df_bike, x="temp_feels_like", kde=True, color="skyblue")
plt.axvline(
    df_bike["temp_feels_like"].mean(),
    ls="--",
    color="red",
    label=f'Durchschnittliche Temperatur: {df_bike["temp_feels_like"].mean():.2f}',
)

# Festlegen des Titels und der Achsenbeschriftungen auf Deutsch
plt.title("Verteilung der gefühlten Temperatur", fontsize=6)
plt.xlabel("Gefühlte Temperatur")
plt.ylabel("Anzahl der Datensätze")

# Anzeigen der Legende und des Diagramms
plt.legend()
plt.show()

plt.figure(figsize=(20, 5), dpi=300)
fig = sns.distplot(df_bike["humidity"])
plt.xlabel("", fontsize=15)
plt.title("Distribution of Humidity", fontsize=20)
sns.despine(top=True)
plt.axvline(df_bike["humidity"].mean(), ls="--")

<matplotlib.lines.Line2D at 0x1fba4fc1650>

plt.figure(figsize=(20, 5), dpi=300)
fig = sns.barplot(data=df_bike, x="time", y="count", palette="deep")
plt.xlabel("Zeit", fontsize=15)
plt.ylabel("Anzahl der Fahrradfahrten", fontsize=15)
plt.title("Verbreitung der Fahrradfahrten nach Tageszeit", fontsize=20)
sns.despine(top=True)
plt.axhline(df_bike["count"].sum() / len(df_bike), ls="--")

<matplotlib.lines.Line2D at 0x1fba7baf210>

plt.figure(figsize=(20, 5), dpi=300)
fig = sns.barplot(data=df_bike, x="time", y="count", hue="is_holiday", palette="deep")
plt.xlabel("Time", fontsize=15)
plt.ylabel("No. of Bike Rides", fontsize=15)
plt.title("Bike rides spread according to weekends/holidays and weekdays", fontsize=20)
sns.despine(top=True)
plt.axhline(df_bike["count"].sum() / len(df_bike), ls="--")

<matplotlib.lines.Line2D at 0x1fba99d9650>

df_bike

plt.figure(figsize=(20, 5), dpi=300)
fig = sns.barplot(data=df_bike, x="month", y="count", palette="deep")
plt.xlabel("Monat", fontsize=15)
plt.ylabel("Anzahl der Fahrradfahrten", fontsize=15)
plt.title("Verteilung der Fahrradfahrten über das Jahr", fontsize=20)

# إزالة الحواف العلوية
sns.despine(top=True)
# إضافة خط للقيمة المتوسطة
plt.axhline(df_bike["count"].sum() / len(df_bike), ls="--")
# إظهار الرسم البياني
plt.show()

plt.figure(figsize=(20, 5), dpi=300)

# إنشاء الرسم البياني
fig = sns.barplot(
    data=df_bike,
    x="day_of_week",
    y="count",
    palette="deep",
    order=[
        "Monday",
        "Tuesday",
        "Wednesday",
        "Thursday",
        "Friday",
        "Saturday",
        "Sunday",
    ],
)

# تعيين التسميات
plt.xlabel("Tag der Woche", fontsize=15)
plt.ylabel("Anzahl der Fahrradfahrten", fontsize=15)
plt.title("Verteilung der Fahrradfahrten über die Woche", fontsize=20)


# إزالة الحواف العلوية
sns.despine(top=True)

# إضافة خط للقيمة المتوسطة
plt.axhline(df_bike["count"].sum() / len(df_bike), ls="--")

# إظهار الرسم البياني
plt.show()

plt.figure(figsize=(20, 5), dpi=300)

# إنشاء الرسم البياني
fig = sns.barplot(
    data=df_bike,
    x="season_name",
    y="count",
    palette="deep",
)

# تعيين التسميات
plt.xlabel("Jahreszeit", fontsize=15)
plt.ylabel("Anzahl der Fahrradfahrten", fontsize=15)
plt.title("Verteilung der Fahrradfahrten nach Jahreszeiten", fontsize=20)

# إزالة الحواف العلوية
sns.despine(top=True)

# إضافة خط للقيمة المتوسطة
plt.axhline(df_bike["count"].sum() / len(df_bike), ls="--")

# إظهار الرسم البياني
plt.show()

plt.figure(figsize=(20, 5), dpi=300)
fig = sns.lineplot(data=df_bike, x="temp_feels_like", y="count")
plt.xlabel("Temperatur", fontsize=15)
plt.ylabel("Anzahl der Fahrradfahrten", fontsize=15)
plt.title("Beziehung zwischen Temperatur und Fahrradfahrten", fontsize=20)
sns.despine(top=True)
plt.axhline(df_bike["count"].sum() / len(df_bike), ls="--")

<matplotlib.lines.Line2D at 0x1fba9b82cd0>

plt.figure(figsize=(20, 5), dpi=300)
fig = sns.lineplot(data=df_bike, x="humidity", y="count")
plt.xlabel("Luftfeuchtigkeit", fontsize=15)
plt.ylabel("Anzahl der Fahrradfahrten", fontsize=15)
plt.title("Beziehung zwischen Luftfeuchtigkeit und Fahrradfahrten", fontsize=20)
sns.despine(top=True)
plt.axhline(df_bike["count"].sum() / len(df_bike), ls="--")

<matplotlib.lines.Line2D at 0x1fba9d31050>

plt.figure(figsize=(20, 5), dpi=300)
fig = sns.lineplot(data=df_bike, x="wind_speed", y="count")
plt.xlabel("Windgeschwindigkeit", fontsize=15)
plt.ylabel("Anzahl der Fahrradfahrten", fontsize=15)
plt.title("Beziehung zwischen Windgeschwindigkeit und Fahrradfahrten", fontsize=20)
sns.despine(top=True)
plt.axhline(df_bike["count"].sum() / len(df_bike), ls="--")

<matplotlib.lines.Line2D at 0x1fba9b21dd0>

df_bike.isna().sum()

timestamp          0
count              0
temp               0
temp_feels_like    0
humidity           0
wind_speed         0
weather_code       0
is_holiday         0
is_weekend         0
season             0
year               0
month              0
time               0
day_of_week        0
season_name        0
day                0
dtype: int64

# إنشاء مؤشر لساعة كل صف
hourly_counts = df_bike[df_bike["count"] > 4000].groupby(df_bike["time"]).size()

# عرض النتائج
print(hourly_counts)

time
07:00:00      2
08:00:00    273
09:00:00      1
13:00:00      3
14:00:00      3
15:00:00      6
16:00:00      5
17:00:00    120
18:00:00     99
19:00:00      2
dtype: int64

df_bike

df_bike[df_bike["count"] > 4000].sort_values(by=["count"]).tail()

# pd.set_option("display.max_rows", None)

share_bikes = df_bike

share_bikes = share_bikes.groupby(
    [
        "day",
        "month",
        "is_holiday",
        "is_weekend",
    ],
    as_index=False,
)["count"].sum()
mean_day = share_bikes["count"].mean()

print("The mean of shared bikes per day is: " + str(mean_day))

The mean of shared bikes per day is: 27268.45479452055

share_bikes[share_bikes["count"] < mean_day][
    ["day", "month", "count", "is_holiday", "is_weekend"]
].sort_values(by=["count"])

df_corr = df_bike.corr(method="pearson", numeric_only=True)
df_corr

cmap = sns.diverging_palette(220, 10, as_cmap=True)
sns.heatmap(
    df_bike.corr(method="pearson", numeric_only=True),
    cmap=cmap,
    vmax=1.0,
    vmin=-1.0,
    center=0,
    square=True,
    linewidths=0.5,
)
plt.title("Correlation between attributes")

Text(0.5, 1.0, 'Correlation between attributes')

df_bike["n_day"] = pd.to_numeric(df_bike["timestamp"].dt.strftime("%d"))
df_bike["n_month"] = df_bike["timestamp"].dt.month
df_bike["n_hour"] = df_bike["timestamp"].dt.hour
df_bike["n_day_of_week"] = df_bike["timestamp"].dt.dayofweek
df_bike["n_day_of_year"] = df_bike["timestamp"].dt.dayofyear
df_bike.tail()

# df_bike["cnt_log"] = np.log1p(df_bike["count"])

""" sns.distplot(df_bike["cnt_log"])
plt.title("Histogram of logarithm of count of shared bikes")
plt.ylabel("Density") """

' sns.distplot(df_bike["cnt_log"])\nplt.title("Histogram of logarithm of count of shared bikes")\nplt.ylabel("Density") '

import matplotlib.pyplot as plt

# تعيين حجم الرسم والدقة
plt.figure(figsize=(6, 3), dpi=100)

# رسم البيانات
plt.plot(
    df_bike["timestamp"],
    df_bike["count"],
    color="orange",
    marker="o",
    linestyle="-",
    linewidth=1,
    markersize=3,
)

plt.title("Anzahl der Fahrräder nach Datum", fontsize=16)
plt.xlabel("Datum", fontsize=12)
plt.ylabel("Geteilte Fahrräder", fontsize=12)
plt.grid(True, linestyle="--", alpha=0.6)
plt.xticks(rotation=45)
plt.tight_layout()
plt.show()

outliers = df_bike[df_bike["count"] > 6000]
outliers

df_bike = df_bike.drop(outliers.index)

# تعيين حجم الرسم والدقة
plt.figure(figsize=(8, 4), dpi=100)

# رسم البيانات
plt.plot(
    df_bike["timestamp"],
    df_bike["count"],
    color="green",
    marker="o",
    linestyle="-",
    linewidth=1,
    markersize=3,
)

plt.title("Anzahl der Fahrräder nach Datum", fontsize=16)
plt.xlabel("Datum", fontsize=12)
plt.ylabel("Geteilte Fahrräder", fontsize=12)
plt.grid(True, linestyle="--", alpha=0.6)
plt.xticks(rotation=45)
plt.tight_layout()
plt.show()

# pd.set_option("display.max_rows", None)

start_date = "2015-12-24"
end_date = "2016-01-05"
filtered_data = df_bike[
    (df_bike["timestamp"] >= start_date) & (df_bike["timestamp"] <= end_date)
]

selected_data = filtered_data[["timestamp", "count"]]
selected_data

df_bike.reset_index(drop=True, inplace=True)

df_bike

columns_to_encode = ["weather_code", "season"]

encoder_weather = OneHotEncoder(sparse=False)
encoder_weather.fit(df_bike[["weather_code"]])
encoded_weather = encoder_weather.transform(df_bike[["weather_code"]])
encoded_weather_df = pd.DataFrame(
    encoded_weather, columns=encoder_weather.get_feature_names_out(["weather_code"])
)
encoded_weather_df

encoder_season = OneHotEncoder(sparse=False)
encoder_season.fit(df_bike[["season"]])
encoded_season = encoder_season.transform(df_bike[["season"]])
encoded_season_df = pd.DataFrame(
    encoded_season, columns=encoder_season.get_feature_names_out(["season"])
)

encoded_season_df

df_bike = pd.concat(
    [df_bike, encoded_weather_df, encoded_season_df],
    axis=1,
)
df_bike

df_bike_modified = df_bike.drop(
    [
        "timestamp",
        "temp_feels_like",
        "weather_code",
        "season",
        "day_of_week",
        "month",
        "season_name",
        "day",
        "time",
        "is_weekend",
        # "count",
    ],
    axis=1,
)
df_bike_modified.head()

df_bike_modified.columns

Index(['count', 'temp', 'humidity', 'wind_speed', 'is_holiday', 'year',
       'n_day', 'n_month', 'n_hour', 'n_day_of_week', 'n_day_of_year',
       'weather_code_1.0', 'weather_code_2.0', 'weather_code_3.0',
       'weather_code_4.0', 'weather_code_7.0', 'weather_code_10.0',
       'weather_code_26.0', 'season_0.0', 'season_1.0', 'season_2.0',
       'season_3.0'],
      dtype='object')

X = df_bike_modified.drop(["count"], axis=1)
y = df_bike_modified["count"]

X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)

scaler = MinMaxScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

def custom_rmse(y_true, y_pred):
    mse = mean_squared_error(y_true, y_pred)
    return abs(mse) ** 0.5

scorers = {
    "MSE": make_scorer(mean_squared_error, greater_is_better=False),
    "MAE": make_scorer(mean_absolute_error, greater_is_better=False),
    "R-squared": make_scorer(
        r2_score, greater_is_better=False
    ),  # Greater is better for R-squared
    "RMSE": make_scorer(custom_rmse, greater_is_better=False),
}

# Define the models to test
models = {
    "LinearRegression": LinearRegression(),
    "RandomForest": RandomForestRegressor(random_state=42),
}

# Initialize a DataFrame to store the results
results = []

# Test each algorithm
for model_name, model in models.items():
    for metric_name, scorer in scorers.items():
        start_time = time.time()

        # Then, estimate the performance
        scores = -cross_val_score(model, X_train_scaled, y_train, cv=5, scoring=scorer)
        end_time = round(time.time() - start_time, 3)

        mean_score = round(scores.mean(), 4)
        std_score = round(scores.std(), 4)

        results.append([model_name, metric_name, mean_score, std_score, end_time])

# Create a DataFrame from the results
columns = ["Algorithm", "Metric", "Mean Score", "Std Score", "Execution Time (s)"]
results_df = pd.DataFrame(results, columns=columns)

# Print the results
print(results_df)

          Algorithm     Metric   Mean Score   Std Score  Execution Time (s)
0  LinearRegression        MSE  786982.4025  45980.1447               0.045
1  LinearRegression        MAE     646.6316     16.3621               0.038
2  LinearRegression  R-squared       0.3201      0.0185               0.041
3  LinearRegression       RMSE     886.7421     25.9023               0.039
4      RandomForest        MSE   39235.4922   1527.3812              87.777
5      RandomForest        MAE     113.6560      1.8043              80.355
6      RandomForest  R-squared       0.9660      0.0020              85.433
7      RandomForest       RMSE     198.0423      3.8400              83.310

scoring = {
    "RMSE": "neg_root_mean_squared_error",
    "MAE": "neg_mean_absolute_error",
    "R2": "r2",
    "Max Error": "max_error",
}
grid_values_lr = {
    "fit_intercept": [True, False],
    "positive": [True, False],  # إضافة معامل positive بدلاً من normalize
}
# قم بإنشاء نموذج الانحدار الخطي
model_lr = LinearRegression()

# قم بإنشاء مثيل GridSearchCV لنموذج الانحدار الخطي
grid_lr = GridSearchCV(
    model_lr, param_grid=grid_values_lr, cv=10, scoring=scoring, refit="RMSE"
)

# نفذ عملية البحث عبر الشبكة
grid_lr.fit(X_train_scaled, y_train)

# احصل على أفضل هيئة لنموذج الانحدار الخطي
best_lr_model = grid_lr.best_estimator_

# عرض أفضل هيئة ونتائج البحث
print("Best Linear Regression Model:")
print("Best Parameters:", grid_lr.best_params_)

# قائمة لتخزين النتائج
results = []

# عرض مقاييس الأداء المتعددة وتخزينها في القائمة
for metric_name, scorer in scoring.items():
    scores = grid_lr.cv_results_[f"mean_test_{metric_name}"]
    mean_score = round(-scores.mean(), 4)
    # std_score = round(scores.std(), 4)
    results.append(
        [
            "LinearRegression",
            metric_name,
            mean_score,
        ]
    )

# عرض النتائج بالشكل المطلوب
for result in results:
    print(result)

Best Linear Regression Model:
Best Parameters: {'fit_intercept': False, 'positive': False}
['LinearRegression', 'RMSE', 909.1428]
['LinearRegression', 'MAE', 671.0793]
['LinearRegression', 'R2', -0.2838]
['LinearRegression', 'Max Error', 4060.4663]

Best RandomForest Model: RandomForestRegressor(bootstrap=False, max_depth=30, max_features='sqrt',
                      n_estimators=750, random_state=42)
Best Parameters: {'random_state': 42, 'n_estimators': 750, 'min_samples_split': 2, 'min_samples_leaf': 1, 'max_features': 'sqrt', 'max_depth': 30, 'bootstrap': False}
Best R-squared Score: 0.8641685995724752
Best RMSE Score: 396.35537684637404

Root Mean Squared Error (RMSE):  393.35467056040477
Mean Absolute Error (MAE):  241.63337230608238
R-squared (Coefficient of Determination):  0.8701924777024996
Max Error:  2627.8388148148147

Root Mean Squared Error (RMSE):  901.1443479611899
Mean Absolute Error (MAE):  660.3401925389475
R-squared (Coefficient of Determination):  0.3187289031567806
Max Error:  4113.699101001589

scoring = {
    "RMSE": "neg_root_mean_squared_error",
    "MAE": "neg_mean_absolute_error",
    "R2": "r2",
    "Max Error": "max_error",
}
grid_values_lr = {
    "fit_intercept": [True, False],
    "positive": [True, False],  # إضافة معامل positive بدلاً من normalize
}
# قم بإنشاء نموذج الانحدار الخطي
model_lr = LinearRegression()

# قم بإنشاء مثيل GridSearchCV لنموذج الانحدار الخطي
grid_lr = GridSearchCV(
    model_lr, param_grid=grid_values_lr, cv=10, scoring=scoring, refit="RMSE"
)

# نفذ عملية البحث عبر الشبكة
grid_lr.fit(X_train_scaled, y_train)

# احصل على أفضل هيئة لنموذج الانحدار الخطي
best_lr_model = grid_lr.best_estimator_

# عرض أفضل هيئة ونتائج البحث
print("Best Linear Regression Model:")
print("Best Parameters:", grid_lr.best_params_)

# قائمة لتخزين النتائج
results = []

# عرض مقاييس الأداء المتعددة وتخزينها في القائمة
for metric_name, scorer in scoring.items():
    scores = grid_lr.cv_results_[f"mean_test_{metric_name}"]
    mean_score = round(-scores.mean(), 4)
    # std_score = round(scores.std(), 4)
    results.append(
        [
            "LinearRegression",
            metric_name,
            mean_score,
        ]
    )

# عرض النتائج بالشكل المطلوب
for result in results:
    print(result)

Best Linear Regression Model:
Best Parameters: {'fit_intercept': False, 'positive': False}
['LinearRegression', 'RMSE', 909.1428]
['LinearRegression', 'MAE', 671.0793]
['LinearRegression', 'R2', -0.2838]
['LinearRegression', 'Max Error', 4060.4663]

from sklearn.model_selection import RandomizedSearchCV


grid_values_rf = {
    "n_estimators": [10, 100, 500, 750, 1000],
    "max_depth": [5, 9, 15, 30, None],
    "min_samples_split": [2, 5, 10],
    "min_samples_leaf": [1, 2, 4],
    "max_features": ["auto", "sqrt"],
    "bootstrap": [True, False],
    "random_state": [42],
}

# قم بإنشاء نموذج RandomForestRegressor
rf_model = RandomForestRegressor()

# إنشاء مثيل RandomizedSearchCV لنموذج RandomForestRegressor باستخدام مقياسي RMSE و R-squared
grid_rf = RandomizedSearchCV(
    rf_model,
    param_distributions=grid_values_rf,
    cv=5,
    scoring=["r2", "neg_root_mean_squared_error"],  # استخدم قائمة المقاييس
    n_jobs=-1,
    n_iter=100,
    refit="neg_root_mean_squared_error",  # grid_rf.best_score_ يعيد أفضل قيمة للمقياس الذي تم تحديده كـ refit عند إنشاء
)

# نفذ عملية البحث عبر الشبكة
grid_rf.fit(X_train_scaled, y_train)

# الحصول على أفضل هيئة لنموذج RandomForestRegressor
best_rf_model = grid_rf.best_estimator_

# عرض أفضل هيئة ونتائج البحث للمقاييس المختارة
print("Best RandomForest Model:", best_rf_model)
print("Best Parameters:", grid_rf.best_params_)
print(
    "Best R-squared Score:", grid_rf.cv_results_["mean_test_r2"][grid_rf.best_index_]
)  # استخدم R-squared
print("Best RMSE Score:", -grid_rf.best_score_)  # استخدم الناتج السلبي للمقياس RMSE

Best RandomForest Model: RandomForestRegressor(bootstrap=False, max_depth=30, max_features='sqrt',
                      n_estimators=750, random_state=42)
Best Parameters: {'random_state': 42, 'n_estimators': 750, 'min_samples_split': 2, 'min_samples_leaf': 1, 'max_features': 'sqrt', 'max_depth': 30, 'bootstrap': False}
Best R-squared Score: 0.8641685995724752
Best RMSE Score: 396.35537684637404

def plot_scores(model, X_test, y_test):
    # التحقق من أن عدد الصفوف في y_test متساوٍ لعدد الصفوف في X_test
    assert y_test.shape[0] == X_test.shape[0]

    predictions = model.predict(X_test)
    actual_values = y_test

    # حساب وعرض خطأ الجذر التربيعي لمتوسط الأخطاء المربعية (RMSE)
    rmse = np.sqrt(mean_squared_error(actual_values, predictions))
    print("Root Mean Squared Error (RMSE): ", rmse)

    # حساب وعرض الخطأ المطلق المتوسط (MAE)
    mae = mean_absolute_error(actual_values, predictions)
    print("Mean Absolute Error (MAE): ", mae)

    # حساب وعرض معامل التحديد R-squared
    r_squared = r2_score(actual_values, predictions)
    print("R-squared (Coefficient of Determination): ", r_squared)

    # حساب وعرض مقياس الخطأ الأقصى (Max Error)
    max_err = max_error(actual_values, predictions)
    print("Max Error: ", max_err)

plot_scores(grid_rf.best_estimator_, X_test_scaled, y_test)

Root Mean Squared Error (RMSE):  393.35467056040477
Mean Absolute Error (MAE):  241.63337230608238
R-squared (Coefficient of Determination):  0.8701924777024996
Max Error:  2627.8388148148147

plot_scores(grid_lr.best_estimator_, X_test_scaled, y_test)

Root Mean Squared Error (RMSE):  901.1443479611899
Mean Absolute Error (MAE):  660.3401925389475
R-squared (Coefficient of Determination):  0.3187289031567806
Max Error:  4113.699101001589

	timestamp	count	temp	temp_feels_like	humidity	wind_speed	weather_code	is_holiday	is_weekend	season	year
count	17414	17414.000000	17414.000000	17414.000000	17414.000000	17414.000000	17414.000000	17414.000000	17414.000000	17414.000000	17414.000000
mean	2016-01-03 22:31:00.571953664	1143.101642	12.468091	11.520836	72.324954	15.913063	2.722752	0.022051	0.285403	1.492075	2015.507810
min	2015-01-04 00:00:00	0.000000	-1.500000	-6.000000	20.500000	0.000000	1.000000	0.000000	0.000000	0.000000	2015.000000
25%	2015-07-04 20:15:00	257.000000	8.000000	6.000000	63.000000	10.000000	1.000000	0.000000	0.000000	0.000000	2015.000000
50%	2016-01-03 15:30:00	844.000000	12.500000	12.500000	74.500000	15.000000	2.000000	0.000000	0.000000	1.000000	2016.000000
75%	2016-07-04 15:45:00	1671.750000	16.000000	16.000000	83.000000	20.500000	3.000000	0.000000	1.000000	2.000000	2016.000000
max	2017-01-03 23:00:00	7860.000000	34.000000	34.000000	100.000000	56.500000	26.000000	1.000000	1.000000	3.000000	2017.000000
std	NaN	1085.108068	5.571818	6.615145	14.313186	7.894570	2.341163	0.146854	0.451619	1.118911	0.508157

	temp	humidity	wind_speed	season	count
timestamp
2015-01-04	2.479167	94.270833	7.500000	Winter	9234
2015-01-11	9.002994	75.008982	22.715569	Winter	129038
2015-01-18	6.863095	73.092262	20.619048	Winter	139559
2015-01-25	3.434524	78.604167	11.922619	Winter	146866
2015-02-01	5.351190	73.446429	21.761905	Winter	133247
...	...	...	...	...	...
2016-12-11	9.886905	87.479167	12.470238	Winter	174112
2016-12-18	9.187500	90.505952	8.985119	Winter	168680
2016-12-25	9.062500	81.613095	17.470238	Winter	151948
2017-01-01	5.169643	88.250000	11.345238	Winter	78188
2017-01-08	3.562500	82.270833	15.052083	Winter	31259

	day	month	count	is_holiday	is_weekend
364	2016-01-03	January	4869	0.0	1.0
727	2017-01-01	January	6421	0.0	1.0
363	2016-01-02	January	7195	0.0	1.0
719	2016-12-24	December	7890	0.0	1.0
448	2016-03-27	March	8123	0.0	1.0
...	...	...	...	...	...
438	2016-03-17	March	27060	0.0	0.0
453	2016-04-01	April	27137	0.0	0.0
197	2015-07-20	July	27163	0.0	0.0
621	2016-09-17	September	27182	0.0	1.0
460	2016-04-08	April	27191	0.0	0.0

	count	temp	temp_feels_like	humidity	wind_speed	weather_code	is_holiday	is_weekend	season	year
count	1.000000	0.388798	0.369035	-0.462901	0.116295	-0.166633	-0.051698	-0.096499	-0.116180	0.010046
temp	0.388798	1.000000	0.988344	-0.447781	0.145471	-0.097114	-0.042233	-0.005342	-0.285851	-0.037959
temp_feels_like	0.369035	0.988344	1.000000	-0.403495	0.088409	-0.098385	-0.040051	-0.008510	-0.285900	-0.044972
humidity	-0.462901	-0.447781	-0.403495	1.000000	-0.287789	0.334750	0.032068	0.028098	0.290381	0.072443
wind_speed	0.116295	0.145471	0.088409	-0.287789	1.000000	0.124803	-0.002606	0.011479	0.010305	-0.094739
weather_code	-0.166633	-0.097114	-0.098385	0.334750	0.124803	1.000000	0.012939	0.042362	0.098976	-0.009234
is_holiday	-0.051698	-0.042233	-0.040051	0.032068	-0.002606	0.012939	1.000000	-0.094898	-0.032488	0.034631
is_weekend	-0.096499	-0.005342	-0.008510	0.028098	0.011479	0.042362	-0.094898	1.000000	0.001067	0.003049
season	-0.116180	-0.285851	-0.285900	0.290381	0.010305	0.098976	-0.032488	0.001067	1.000000	0.024400
year	0.010046	-0.037959	-0.044972	0.072443	-0.094739	-0.009234	0.034631	0.003049	0.024400	1.000000

	timestamp	count	temp	temp_feels_like	humidity	wind_speed	weather_code	season	...	month	time	day_of_week	season_name	day	n_day	n_month	n_hour	n_day_of_week	n_day_of_year
4461	2015-07-09 08:00:00	7531	14.5	14.5	61.0	19.0	2.0	1.0	...	July	08:00:00	Thursday	Summer	2015-07-09	9	7	8	3	190
4469	2015-07-09 16:00:00	6033	23.0	22.0	26.0	11.0	1.0	1.0	...	July	16:00:00	Thursday	Summer	2015-07-09	9	7	16	3	190
4470	2015-07-09 17:00:00	7860	23.0	22.0	27.0	11.0	1.0	1.0	...	July	17:00:00	Thursday	Summer	2015-07-09	9	7	17	3	190
4471	2015-07-09 18:00:00	6913	22.5	21.5	29.0	13.0	1.0	1.0	...	July	18:00:00	Thursday	Summer	2015-07-09	9	7	18	3	190
5129	2015-08-06 08:00:00	6585	19.0	19.0	78.0	12.0	7.0	1.0	...	August	08:00:00	Thursday	Summer	2015-08-06	6	8	8	3	218
5138	2015-08-06 17:00:00	7208	22.5	22.5	55.0	17.5	2.0	1.0	...	August	17:00:00	Thursday	Summer	2015-08-06	6	8	17	3	218
5139	2015-08-06 18:00:00	6394	21.5	21.5	58.5	20.0	2.0	1.0	...	August	18:00:00	Thursday	Summer	2015-08-06	6	8	18	3	218

London bike Machine Learning project¶

Section 1. Daten kennenlernen¶

1.2 Data Cleaning¶

Section 2. Data visualization¶

Korrelationen in Wetterdaten:¶

Korrelationen zwischen der Nutzung von Fahrrädern und Wetterdaten:¶

Section 3. Data analysis¶

3.1 Correlation between attributes¶

Abschnitt 4. Datenverarbeitung¶

4.2. Removing outliers¶

4.3. One-hot encoding¶

4.4 Entfernen redundanter Daten¶

Section 5. Vorhersagemodelle¶

5.1 Statistical Models¶

5.1.1 Linear Regression + RandomForest Model¶

5.1.2 Optimization of parameters (Grid search)¶

Section 6. Results¶

	timestamp	cnt	t1	t2	hum	wind_speed	weather_code	is_weekend	season
0	2015-01-04 00:00:00	182	3.0	2.0	93.0	6.0	3.0	1.0	3.0
1	2015-01-04 01:00:00	138	3.0	2.5	93.0	5.0	1.0	1.0	3.0
2	2015-01-04 02:00:00	134	2.5	2.5	96.5	0.0	1.0	1.0	3.0
3	2015-01-04 03:00:00	72	2.0	2.0	100.0	0.0	1.0	1.0	3.0
4	2015-01-04 04:00:00	47	2.0	0.0	93.0	6.5	1.0	1.0	3.0

	timestamp	count	temp	temp_feels_like	humidity	wind_speed	weather_code	is_holiday	is_weekend	season	year	month	time	day_of_week	season_name	day
0	2015-01-04 00:00:00	182	3.0	2.0	93.0	6.0	3.0	0.0	1.0	3.0	2015	January	00:00:00	Sunday	Winter	2015-01-04
1	2015-01-04 01:00:00	138	3.0	2.5	93.0	5.0	1.0	0.0	1.0	3.0	2015	January	01:00:00	Sunday	Winter	2015-01-04
2	2015-01-04 02:00:00	134	2.5	2.5	96.5	0.0	1.0	0.0	1.0	3.0	2015	January	02:00:00	Sunday	Winter	2015-01-04
3	2015-01-04 03:00:00	72	2.0	2.0	100.0	0.0	1.0	0.0	1.0	3.0	2015	January	03:00:00	Sunday	Winter	2015-01-04
4	2015-01-04 04:00:00	47	2.0	0.0	93.0	6.5	1.0	0.0	1.0	3.0	2015	January	04:00:00	Sunday	Winter	2015-01-04
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
17409	2017-01-03 19:00:00	1042	5.0	1.0	81.0	19.0	3.0	0.0	0.0	3.0	2017	January	19:00:00	Tuesday	Winter	2017-01-03
17410	2017-01-03 20:00:00	541	5.0	1.0	81.0	21.0	4.0	0.0	0.0	3.0	2017	January	20:00:00	Tuesday	Winter	2017-01-03
17411	2017-01-03 21:00:00	337	5.5	1.5	78.5	24.0	4.0	0.0	0.0	3.0	2017	January	21:00:00	Tuesday	Winter	2017-01-03
17412	2017-01-03 22:00:00	224	5.5	1.5	76.0	23.0	4.0	0.0	0.0	3.0	2017	January	22:00:00	Tuesday	Winter	2017-01-03
17413	2017-01-03 23:00:00	139	5.0	1.0	76.0	22.0	2.0	0.0	0.0	3.0	2017	January	23:00:00	Tuesday	Winter	2017-01-03

	timestamp	count
8451	2015-12-24 00:00:00	175
8452	2015-12-24 01:00:00	159
8453	2015-12-24 02:00:00	104
8454	2015-12-24 03:00:00	74
8455	2015-12-24 04:00:00	66
...	...	...
8735	2016-01-04 20:00:00	614
8736	2016-01-04 21:00:00	397
8737	2016-01-04 22:00:00	293
8738	2016-01-04 23:00:00	174
8739	2016-01-05 00:00:00	93

	weather_code_1.0	weather_code_2.0	weather_code_3.0	weather_code_4.0	weather_code_7.0	weather_code_10.0	weather_code_26.0
0	0.0	0.0	1.0	0.0	0.0	0.0	0.0
1	1.0	0.0	0.0	0.0	0.0	0.0	0.0
2	1.0	0.0	0.0	0.0	0.0	0.0	0.0
3	1.0	0.0	0.0	0.0	0.0	0.0	0.0
4	1.0	0.0	0.0	0.0	0.0	0.0	0.0
...	...	...	...	...	...	...	...
17402	0.0	0.0	1.0	0.0	0.0	0.0	0.0
17403	0.0	0.0	0.0	1.0	0.0	0.0	0.0
17404	0.0	0.0	0.0	1.0	0.0	0.0	0.0
17405	0.0	0.0	0.0	1.0	0.0	0.0	0.0
17406	0.0	1.0	0.0	0.0	0.0	0.0	0.0